An Overview of New Integrated Geometallurgical Research

S G Walters1

ABSTRACT DEVELOPMENT OF NEW GEOMETALLURGICAL

The emerging discipline of ‘geometallurgy’ is not new but is becoming RESEARCH INITIATIVES
increasingly recognised as a discrete and high-value activity that reflects Industry recognition of the need for large-scale, integrated
an ongoing commercial and cultural trend towards more effective mine
site integration and optimisation. The AMIRA International P843
geometallurgical research resulted in commencement of a major
‘GeMIII’ project (Geometallurgical Mapping and Mine Modelling) is a new initiative in mid-2005. The AMIRA International
major new cross discipline research collaboration designed to provide P843 ‘GeMIII’ project (Geometallurgical Mapping and Mine
support for this emerging industry trend. Modelling) is a cross discipline collaboration that brings together
The project is currently supported by 18 global mineral companies and three major Australian research groups for the first time; CODES
involves collaboration between the CODES Centre of Excellence the ARC Centre of Excellence in Ore Deposits at the University
(University of Tasmania), JKMRC and BRC (University of Queensland) of Tasmania, a world-leader in economic geology-related
and CSIRO Exploration and Mining. These groups represent recognised research; the Julius Kruttschnitt Mineral Research Centre
research excellence in the fields of economic geology, automated core (JKMRC) at the University of Queensland, a world-leader in
logging, advanced microscopy and process mineralogy, mineral mining and mineral processing research; and the WH Bryan
processing, mine planning and resource optimisation. Mining Geology Research Centre (BRC) at the University of
The needs and challenges of geometallurgy as an integrated activity Queensland, which aims to be a world leader in mining
will be outlined together with the ongoing contribution and aims of the geostatistics, operations research and optimisation in mine
AMIRA P843 ‘GeMIII’ project. The main emphasis is on development of design and planning.
predictive geometallurgical indices for key processing performance
parameters that reflect inherent geological variability. The project aims to deliver fundamental knowledge, tools and
The ultimate aim is to provide predictive indices of geometallurgical methods for more effective approaches to geometallurgical
performance that can be confidently embedded into resource models and definition of mineralogy, element deportment and texture that
used for risk-defined mine planning and optimisation. can be linked to mineral processing performance and efficiency
at a range of scales. Outcomes will be improved indices of
processing performance that can be embedded in resource block
INTRODUCTION models to define processing domains, which can be exploited in
The emerging discipline of ‘geometallurgy’ is not new but is mine planning and optimisation.
becoming increasingly recognised as a discrete and high-value To achieve these aims the project is involved in a range of
activity that reflects an ongoing commercial and cultural trend technology developments that include more automated core
towards more effective mine site integration and optimisation. logging technologies, new applications of automated microscopy,
Geometallurgy involves a quantified and spatially constrained and development of software-based texture analysis and
approach to ore characterisation in terms of relationships to categorisation techniques. Attributes derived from these
critical processing performance behaviours (Williams and measurement technologies are used to develop predictive
Richardson, 2004). processing models linked to new small-scale physical testing
regimes (Walters and Kojovic, 2006).
For base and precious metal mining operations these
behaviours include blasting, crushing, grinding, liberation and The AMIRA P843 ‘GeMIII’ project is currently supported by
recovery. Related issues include tracking deportment of 21 companies including BHP Billiton, Rio Tinto, Vale and Anglo
deleterious elements and minerals; providing inputs into American. Project funding is in excess of A$8 million supporting
environmental waste rock management; and increasing a team of over 30 researchers and support staff.
comminution energy efficiency. A key outcome is enhanced
economic optimisation of mining. WHAT IS GEOMETALLURGY?
Geometallurgy is a cross-discipline challenge that requires Geometallurgy complements but does not replace existing
more effective communication, shared methodologies and a approaches to design and optimisation of mining and mineral
common technical language between the wide range of processing operations. The primary aim of geometallurgy is to
professional disciplines that support major mining operations – provide constrained inputs that reflect inherent geological
in particular economic geologists, mining engineers and mineral variability and its impact on metallurgical performance. This
processors. Significant disjoints can occur across this shared requires quantification of ore deposits in terms of machine-based
challenge, which are often exacerbated by separate tertiary process parameters such as hardness, comminution energy, size
education pathways. Divisional organisational structures on many reduction, liberation potential and product recovery. These
mine sites can also dilute the ‘one-mine’ strategy required for parameters are used to populate deposit-scale block models
successful geometallurgical implementation. typically through the application of geostatistics. Incorporation
The historical impediments to geometallurgical integration of geometallurgical parameters into resource modelling
require major new education and teaching initiatives backed by supplements traditional geology and grade-based attributes,
new technology developments and coordinated research. enabling a more holistic approach to economic optimisation of
mineral production.
Geometallurgy requires integration across a wide range of
III existing activities and can also be referred to under a variety of
1. Professor, GeM Group, Centre of Excellence in Ore Deposits
(CODES), University of Tasmania, Private Bag 79, Hobart Tas 7000. different descriptors. It includes aspects of process mineralogy,
Julius Kruttschnitt Mineral Research Centre (JKMRC), The mine geology, metallurgy, process control, resource modelling
University of Queensland, Isles Road, Indooroopilly Qld 4068. and geostatistics and can also be referred to under different terms
Email: steve.walters@utas.edu.au such as ‘orebody knowledge’ or ‘ore characterisation’.

Ninth International Congress for Applied Mineralogy Brisbane, QLD, 8 - 10 September 2008 79
S G WALTERS

Detailed geometallurgical models that reflect inherent implications for overall sustainability and reduction of energy
variability aim to reduce technical risk associated with the design and environmental footprints.
and operation of mines. A systematic geometallurgical approach One of the key underlying themes behind the current
is particularly important in feasibility as a precursor to design, emergence of geometallurgy is a move towards low-cost physical
where physical access to and knowledge of an ore deposit is testing, which can be applied to small sample volumes that are
typically limited and evolving. Geometallurgical information can suitable for defining natural variability. For geometallurgical
be used to optimise flow sheet design and equipment sizing with modelling large data sets related to small sample volumes is a
a view to predicting and optimising plant performance and much more effective statistical approach to defining natural
production over the life of the project. variability than a small number of ‘more precise’ data points. The
Constrained sampling that reflects and defines inherent end result of this type of approach is a multitiered sampling and
orebody variability is a key geometallurgical requirement. There testing strategy, with large numbers of relatively low-cost
are a number of factors and assumptions in current practice ‘comparative’ tests used to define variability followed by small
which can hinder this process and in some cases produce numbers of high precision ‘bankable’ tests representative
seriously flawed outcomes. of variability.
Many of the traditional geological parameters used to describe Even with larger numbers of more spatially representative
ore systems, such as primary lithology or alteration styles, are tests, these outcomes still represent sparse data compared to the
typically not calibrated with or related to processing parameters. overall volume of an ore deposit. There are many examples in
Application of predetermined geological boundaries to current practice of a few tens of processing performance data
geometallurgical sampling and data modelling has the potential points being used to populate an entire block model. Processing
to introduce serious bias and disguise natural populations and performance data is generally two to three orders of magnitude
trends in processing parameters. The definition of ‘ore type’ in less abundant than grade data with no guarantee that it behaves
current practice typically involves intersection of lithology and in a linear or additive fashion suitable for application of
grade envelopes, with both attributes an imperfect proxy for geostatistics.
processing performance. For example, grade that is not expressed The trend towards larger numbers of more representative
or modified in terms of constituent mineralogy (eg specific Cu geometallurgical samples partly overcomes some of these
species) and ultimately recovery is not an effective problems, particularly with respect to the potential assessment of
geometallurgical attribute. additivity and linearity through the use of cross-validation.
There is also a tendency to composite core-based sampling to However, effective extrapolation of geometallurgical processing
represent scale-up to production or to satisfy logistical performance attributes to the deposit scale requires a suite of
requirements of physical testing procedures. Production scale-up statistical and population modelling techniques in addition to
typically involves matching a projected feed grade or classical geostatistics.
representation of a planned mining block. In the case of physical
testing large sample volume requirements (eg >100 kg) partly OVERVIEW OF AMIRA P843 GEMIII RESEARCH
reflects a desire to meaningfully subsample particulate material PROJECT
and size fractions during testing. However, this approach fails to
acknowledge the significant problems large non-representative A key aspect of the AMIRA P843 GeMIII research project is the
and unconstrained sample compositing can introduce. Precise development of a suite of new or modified measurement and
test results with uncertain representivity are a common outcome testing methodologies that can be used for early-stage
in current practice. geometallurgical characterisation typically at core scale during
Composited samples tend to disguise natural variability. Once feasibility. The aim is to provide more automated ‘machine
samples are composited for destructive testing, understanding vision’ capabilities to supplement or replace visual core logging
(and potentially exploiting) this variability becomes difficult. approaches; develop more flexible and cost-effective capabilities
The desire to introduce compositing early into sampling for automated mineral mapping and microscopy; and deliver a
campaigns reflects two main drivers. Many physical tests for suite of integrated small-scale and lower-cost physical tests for
comminution and flotation are expensive, resulting in relatively measuring processing performance parameters. This approach
few samples within constrained budgets. In this case composites must be capable of application to shared sample volumes ideally
are often regarded as increasing the ‘influence’ of results from at the scale of geochemical assays.
limited sample sets. However, typical outcomes are small data Interpretation and modelling of much larger data sets with
sets with a high degree of statistical noise that make it difficult to more extensive geometallurgically-relevant attributes requires a
define natural populations. range of associated software-based tools. Within the GeM project
The other main driver is a desire to scale-up sampling to this has involved modification of existing software and
represent ‘average’ feed within a projected mine plan. The development of new software capabilities.
ultimate engineering solution to inherent variability is delivery of
steady state ‘average’ feed over the life of the mining operation Automated core logging
through a process of blending. This is difficult to achieve, Much of the non-grade information derived from drill core is a
particularly in many metalliferous ore deposits, but can appear to function of visual inspection as part of geological and
be a desirable outcome from an engineering perspective. geotechnical logging. Industry has clearly identified this process
It is important to know what processing behaviour options are as a major weak link in more quantified geometallurgical
present in an ore deposit before a final design is chosen. In applications.
addition, ‘average’ ore feed typically does not represent a simple Visual logging is generally highly subjective, experience-based,
linear combination of its constituent parts and does not often conducted by junior personnel and difficult to support given
necessarily produce optimal economic outcomes. the current trend to de-skilling in the industry. Many of the
Effective identification of constituent components of an ore geological concepts and observations used in routine logging are
deposit through application of geometallurgy provides the typically related to exploration aspects and are not calibrated
opportunity to increase efficiency and optimisation of mining with or validated against processing performance. However,
and processing. This is particularly important for larger mining many of the interpretative lithologies introduced during routine
operations where there is increasing potential to exploit geological logging have a tendency when combined with grade
variability through multiple processing circuits. There are also to become regarded as ‘ore types’.

80 Brisbane, QLD, 8 - 10 September 2008 Ninth International Congress for Applied Mineralogy
 AN OVERVIEW OF NEW INTEGRATED GEOMETALLURGICAL RESEARCH

The GeM project is employing a novel approach designed to The approach is further enhanced by use of high resolution
help overcome these problems based on automated bench-scale automated imaging systems deployed on GEOTEK core logger
logging instrumentation for petrophysical and mineralogical platforms. Classification of this imaging provides information on
measurement. meso-textural associations and first-pass mineralogy.
Two automated logging systems are currently being used. The
GEOTEK Multi Sensor Core Logger (MSCL) is a bench scale Textural classification and modelling
system based on automated linear feed of drill core past a sensor
array. Within the current project this array is dominated by The integrated combination of more automated core-scale
petrophysical sensors that include gamma attenuation density, imaging together with MLA and optical microscopy generates
P-wave velocity, resistivity and magnetic susceptibility. The large volumes of mineral mapping and texture analysis.
approach is a logical derivate of downhole geophysical logging Development of more efficient software-based tools for handling
that involves bringing rock to the sensor rather than placing the large volumes of increasingly complex mineralogical data is an
sensor into the rock. It allows use of the extensive archived core important aspect of GeM.
available at most sites, which is typically associated with The aim is to provide more effective software-based methods
historical processing performance information. for textural analysis, feature extraction and categorisation with
The majority of petrophysical attributes can provide proxies relevance to processing performance. Typical parameters include
for potential mineral processing performance such as derivates of systematic grain size, grain shape and association from analysis
P-wave velocity and density to strength and elastic moduli. of automated microscopy products (Hunt et al, 2008). Much of
Detailed deposit-based petrophysical characterisation also has this work involves in-house software development. Significant
implications for exploration applications such as constrained advances are being made, for example, in the application of
geophysical inversion modelling. wavelet functions to texture-based image analysis and automated
feature extraction (Leigh, 2008).
Over 80 GEOTEK systems are in use primarily for ocean
drilling and petroleum applications (Schultheiss et al, 2004). Another aspect is the use of classified mineral maps of intact
Although based on mature technology this is the first system texture to simulate and predict potential processing performance.
adapted to the needs of the metalliferous mining industry. The This involves use of sophisticated finite element modelling
logger is fully containerised and has been deployed at sponsor techniques to simulate fundamental mineral-based fracturing and
sites. The GEOTEK logger is also equipped with a high liberation. The work is providing important insights into how
resolution digital imaging system. rock textures behave and interact under processing conditions.
More direct core-based mineralogical information is provided
through technology collaboration involving the CSIRO Small-scale physical testing
HyLoggerTM (Huntington et al, 2006). HyLoggerTM is an A wide variety of physical tests are currently employed to assess
automated spectral analysis and imaging system that collects processing performance. These include strength (eg UCS, point
systematic short wave length infrared (SWIR) reflectance load test), hardness (eg Bond work index) and comminution (eg
spectra. These can be classified into dominant SWIR-responsive JK drop weight and sag mill comminution) together with
mineral species and their compositional variations (chlorites, bench-scale flotation or leaching.
white micas, clays, etc). Distribution of ‘soft’ SWIR-responsive The majority of physical testing is conducted during feasibility
phases show relationships to comminution behaviour related to using drill core. Incomplete knowledge or non-representative
decrease of bulk strength. Work is also in progress to evaluate the results can have a major negative effect on achieving design
use of new CSIRO TIR-based logging technologies. capacity and economic performance particularly during start up.
Many current physical tests are based on large sample volumes
Integrated meso- and microscale mineral that require large-scale compositing of core samples. Combined
mapping and microscopy with high cost, this limits sampling density and often results in
In the last decade there have been major ongoing advances in the inappropriate compositing and blending. The ‘noise’ related to
field of automated microanalytical mineral mapping mainly poorly constrained compositing typically generates poor
related to the development of automated SEM-based X-ray comparative statistics based on small sample sets that disguise
microscopy such as the JKTech MLA and Intellection natural geological variability. The aim is to create a new
QEMSCAN® systems. approach to geometallurgical characterisation that enables highly
constrained samples with defined textural and mineralogical
The mineral identification and mapping capabilities provided attributes to be tested for processing performance. This has
by these systems result in quantified digital imaging of classified required development of a suite of new and modified physical
mineral maps amenable to software-based image analysis of testing methods capable of providing meaningful processing
mineralogy and texture (Gu, 2003; Gottlieb et al, 2000). The parameters based on large numbers of small sample volumes
techniques have proven highly effective for the quantified (Walters and Kojovic, 2006).
analysis and interpretation of liberation in comminuted products
such as concentrates and tailings. The majority of major mineral This approach allows fundamental relationships to be
companies now operate in-house facilities. The GeM project is determined based on geologically constrained inputs. The aim is
supported by a dedicated MLA facility at the University of to enable prediction of processing performance through
Tasmania based on an FEI Quanta 600 SEM and access to application of integrated modelling based on textural and
extensive MLA facilities and development at JKMRC. mineralogical attributes.
The lessons of automated SEM-based microscopy are being
used to develop complimentary optical microscopy systems. Multivariate analysis and modelling
Significant advances have occurred in automated optical The diverse set of attributes provided by the core logging,
microscopy mainly driven by the needs of other sectors such as mineral mapping, textural analysis and small-scale testing suites
pathology. Two Leica DM600M computer-controlled optical in GeM provides a large and potentially complex multivariate
microscopy systems are being used within the project. In data matrix. This matrix includes intrinsic rock-based properties
combination with application of sophisticated image analysis such as detailed mineralogy and texture, combined with
software, this provides routine texture-based digital image capture machine-based testing outcomes that can be used as proxies for
and classification (Berry, Walters and McMahon, 2008). processing performance.

Ninth International Congress for Applied Mineralogy Brisbane, QLD, 8 - 10 September 2008 81
S G WALTERS

Given sufficient sampling density this type of matrix can acknowledges financial support and permission to publish from
be used to investigate inherent geological variability and sponsors of the AMIRA International P843 GEMIII Project –
define meaningful population trends. Attributes related to Anglo Gold Ashanti, Anglo Platinum, Barrick, BHP Billiton,
processing performance can be used to identify and constrain Codelco, Vale Inco, Datamine, Golder Associates, GEOTEK,
geometallurgically-relevant groupings using a range of statistical ioGlobal, Metso Minerals, Newcrest, Newmont, Oxiana, Peñoles,
techniques. Spatial stability of these groupings is the basis of Rio Tinto, Teck Cominco, Vale, Xstrata Copper and Zinifex.
defining processing domain rather than ‘ore types’.
Within GeM predefined geological categories such as rock REFERENCES
type or alteration that have not been calibrated against processing
performance are not used for processing domain definition. Berry, R, Walters, S G and McMahon, C, 2008. Automated mineral
identification by optical microscopy, in Proceedings Ninth
These categories are typically defined visually and are often International Congress for Applied Mineralogy (ICAM), pp 91-94
inherited from early geological assessment with a strong (The Australasian Institute of Mining and Metallurgy: Melbourne).
exploration focus. Gottlieb, P, Wilkie, G, Sutherland, D, Ho-Tun, E, Suthers, S, Perera, K,
The ultimate aim is to produce processing domains for aspects Jenkins, B, Spencer, S, Butcher, A and Rayner, J, 2000. Using
such as throughput or recovery that are independent volumes, quantitative electron microscopy for process mineralogy applications,
which may or may not share common geological boundaries. Journal of Mining, April, pp 24-25.
This is particularly important for large bulk tonnage systems Gu, Y, 2003. Automated scanning electron microscope based mineral
such as porphyry Cu-Au-Mo deposits that typically show liberation analysis – An introduction to the JKMRC/FEI mineral
gradational boundary conditions. While this type of liberation analyser, J Minerals Materials Characterization and
Engineering, 2(1):33-41.
geometallurgical modelling approach can challenge traditional
geological and resource definition thinking, the move towards Hunt, J A, Berry, R, Walters, S G, Bonnici, N, Kamenetsky, M, Nguyen,
K and Evans, C L, 2008. A new look at mineral maps and the
truly integrated geometallurgical modelling and resource potential relationships of extracted data to mineral processing
optimisation requires a significant step change. behaviours, in Proceedings Ninth International Congress for Applied
Mineralogy (ICAM), pp 429-432 (The Australasian Institute of
CONCLUSIONS Mining and Metallurgy: Melbourne).
Huntington, J, Quigley, M, Yang, K, Roache, T, Young, C, Roberts, I,
The emerging area of geometallurgy represents a cross-discipline Whitbourn, I L and Mason, P, 2006. A geological overview of
challenge to the global minerals industry which is driving new HyLogging 18 000 m of core from the Eastern Goldfields of Western
research initiatives. The AMIRA P843 GeMIII project represents Australia, in Proceedings Sixth International Mining Geology
a major integrated response to this challenge. The project involves Conference, pp 45-50 (The Australasian Institute of Mining and
research groups that, although recognised leaders in their Metallurgy: Melbourne).
respective fields, have not previously engaged in collaborative Leigh, G M, 2008. Automatic ore texture analysis for process mineralogy,
in Proceedings Ninth International Congress for Applied Mineralogy
research in this area. The initial focus of the AMIRA P843
(ICAM), pp 433-436 (The Australasian Institute of Mining and
project has been on integrated method development to facilitate Metallurgy: Melbourne)
more cost-effective analysis of small-scale samples and address Schultheiss, P J, Holland, M E, Francis, T J, Roberts, J A and Carter, R
issues of inherent geological variability. This is increasingly M, 2004. Fulfilling the promise of the DSDP/ODP legacy with
moving into developing new approaches to spatial multiparameter logging of archive cores, Eos Trans Am Geophys
geometallurgical modelling. The ultimate aim is to provide Union, 85(47).
predictive indices of geometallurgical performance that can be Walters, S and Kojovic, T, 2006. Geometallurgical mapping and mine
confidently embedded into resource models and used for mine modelling (GeMIII) – The way of the future, in Proceedings
planning and optimisation. SAG2006 Conference, Vancouver, vol IV, pp 411-425.
Williams, S R and Richardson, J M, 2004. Geometallurgical mapping: A
new approach that reduces technical risk, in Proceedings 36th
ACKNOWLEDGEMENTS Annual Meeting Canadian Mineral Processors, pp 241-268.
This research is part of a major collaborative geometallurgical
project being undertaken at CODES (University of Tasmania),
and JKMRC and BRC (University of Queensland). The author

82 Brisbane, QLD, 8 - 10 September 2008 Ninth International Congress for Applied Mineralogy